University of Virginia Library

II

In general, analytical bibliography attempts to resolve dating problems by establishing the location of a given book in the printing schedule of the books produced in a shop during the proximate temporal period. Given the prevalence of shared printing (i.e., a given book printed in sections in several shops), the first order of business is to verify that a book was printed either entirely or in part in the shop specified in the imprint.[5] A combination of ornamental evidence and typographical evidence leaves no doubt that Thomas Orwin (London printer, 1587-1593) printed both Complaints and Daphnaida as well as two other texts published by William Ponsonby (London publisher, 1577-1604) in early 1591.[6] The attempt to determine the position of

the target book in the shop's production schedule relies upon a combination of external evidence (e.g., a Stationers' Register entry date, and/or dated dedications, letters, poems, etc.) and internal evidence consisting of two kinds of bibliographical facts that can be gleaned from an examination of originals.

First, an analysis of the progressive states of contamination of a font (or fonts) used in the target book and other books printed during the proximate period can provide conclusive evidence of the printing sequence.[7] In this instance, a font analysis is not expedient because Orwin used black letter, 94mm roman, and pica roman alternatively as text-fonts during 1591. In combination with the short length of many of the twenty-nine books from 1591, this fact diminishes the available typographical samples of each font to far below the level needed for a reliable progressive-state analysis.

Second, the papers used in books during the proximate period can be surveyed and catalogued with the objective of determining whether any temporal relationships can be discerned which either confirm or disprove the implications of the external dating evidence. The availability of originals has been a limiting factor in regard to the evidence gathered in this survey of papers. Ideally, such a survey includes at least one exemplar of each signature, or “edition-sheet,” of each book printed during the proximate period, and multiple exemplars of many. The Huntington Library possesses only seventeen of the twenty-nine books printed by Orwin in 1591. Because of time restrictions, no attempt was made to examine several of the other twelve books at the Folger Library, where five additional copies of Complaints were examined. Hence, the sample covered in this survey is incomplete in the context of the ideal of examining the papers in every book printed during the proximate period. However, this fact does not detract from the validity

of the present study because its purpose is testing a positive hypothesis, that is: the dedicatory date “this first of Januarie. 1591.” means the calendar year 1591 according to New Style dating. In practical terms, the hypothesis can be confirmed by demonstrating that exactly the same papers were used in both Daphnaida and Complaints. Thus the evidence found in the limited sample permits verification of the New Style date of Daphnaida. [8] A brief discussion of the evidence-inference relationship in the inductive method in the context of the uses of watermark evidence will clarfy this point.

In the inductive method, the sign (i.e., positive, negative) and scope (or extent) of an inference is dependent upon the completeness of the sample from which the evidence is taken. In general, negative inferences are valid only in one circumstance—that the sample contains all the evidence that is relevant, in which case an inference is elevated to the level of a demonstrated conclusion. This limits the scope of a negative inference to statements about the sample itself and nothing more. In other words, the fact that paper A, found in other books of the proximate period, is not found in the target book can only support the inference “paper A does not appear in the target book.” Moreover, the inference is limited only to this copy of the edition and cannot be generalized to “paper A was not used in the printing of the edition.” In regard to inferences based upon evidence of papers (and typography as well), the vagaries of survival always eliminate the possibility of drawing negative inferences about a book or sequence of books. The extant samples of papers are always minuscule in proportion to the actual number of printed sheets in an edition: in the case of Complaints, for example, about sixty-nine whole copies are extant of an edition of unknown size, but undoubtedly consisting of five hundred or more copies. Even if the papers found in all sixty-nine copies were catalogued, it would be invalid to infer negatively that no other papers were used in the printing of the book. The discovery of a seventieth copy, or similarly, an examination of the many incomplete copies, could yield some entirely different watermarks. In this survey, for example, Folger STC23078 copy 3 revealed watermarks not seen in the Huntington copy and Folger copies 1 and 2. Similarly, if the books printed by Orwin in the period of January-April 1592 New Style were surveyed and none of the papers of Complaints were found, it would be invalid to infer that these papers were not used in that period one year later. In short, it can never be inferred that a given paper was not used in a book or in books of a period until every sheet that went thru the press has been examined—an impossible condition which clarifies the inherent limitation of the inductive method in regard to negative inferences.

In contrast, positive inferences can be drawn when a single book or several books contain(s) the evidence that is necessary to demonstrate the truth of a positive hypothesis such as is stated above in regard to Complaints and Daphnaida. In this case, the appearance of exactly the same papers in both books demonstrates temporally proximate printing operations. This general principle arises from our knowledge about the paper supply system and about the various patterns of distribution of papers in books printed at roughly the same time in one shop, or in two or more shops in the case of shared printing. In regard to the distribution of papers in approximately

contemporary books, the processing of papers in the printing house is of special significance.

At paper collection and distribution centers on the continent, bulk lots of paper, either produced at a large paper factory or gathered from thousands of one-vat family operations in northern France, were assembled into quires of 24 or 25 sheets, then into reams of 20 quires containing 480 or 500 sheets respectively, of which 432 or 450 sheets were “good sheets”—the two end quires, or “chording quires,” were made up of defective sheets and took the brunt of the damage from the cords that were tied around the ream to hold it together. The reams were then assembled into bales, the final packaging unit for shipment to England. Paper merchants in London disassembled the bales into reams for sale and delivery to printers as job-lots which usually reflected the composition of the bales. However, instances are known in which the publisher (or printer) purchased a job-lot consisting of two different qualities, types, or sizes of paper with the objective of producing an edition consisting of two qualities of book, e.g., presentation copies, copies intended for marketing in expensive bindings, and copies for general marketing in tied-up packages. After delivery to the printer, the warehouse keeper combined the papers into the heap for use at the press during the next day's printing. In general, for economic reasons, printers ordered job-lots as the need arose rather than maintaining a standing stock of paper— the cost of the latter was far too prohibitive.[9] Throughout the process, batches of papers tended to remain together. As a result, the appearance of exactly the same paper or papers in two different editions (whether of the same work or of different works) constitutes evidence of the highest order of probability that the two editions were printed from the same job-lot of paper.

In theory, two factors limit such inferences merely to the highest order of probability, that is, just short of absolute proof. During the packaging and unpackaging processes, a dissimilar sheet or quire or ream could have been combined with the predominant papers eventually found in a book. Similarly, given the variation in sizes of editions and numbers of sheets in each edition,

the job-lots of paper (delivered to the printer in lots consisting of reams) required for one book or each of a sequence of books did not necessarily match exactly the number of reams found in a bale or sequence of bales from the continent. In such a situation, the merchant would include reams from a different bale or continental shipment to fill out a given order. This appears to have happened in the job-lot delivered to Orwin for Complaints where three “mixtures” of two kinds of papers appear (see later discussion). Finally, a printer could order papers for a work period without regard to the number of different books being machined then, e.g., a sufficient stock of papers to keep two presses at full operation for one week on two or more books in concurrent production.[10]

Similarly, the processing of papers in the printing house could introduce dissimilar papers from another job-lot into a book.[11] Although the “usual” edition sizes mentioned by bibliographers, e.g., 500, 1000, 1250, 1500, do not work out to whole numbers of reams—for example, paper supplied from northern France usually consistend of reams of 450 “good” sheets—the warehouse keeper built the heap in “tokens” of 250 sheets consisting of ten quires, the hourly unit of production at the press.[12] The warehouse keeper had two options for producing the daily heap from the reams included in a job-lot of papers. Given the right correspondence between edition-size, the number of gatherings, and the number of reams in the job-lot, he could sequentially redistribute the 450-sheet reams into an even number of tokens over a sequence of days, and eventually exhaust the job-lot by the end of the printing of a book.[13] Otherwise, he would work the “left- overs” into the early heaps

for the next printing job. Alternately, he could set aside one ream from which to draw the number of quires needed to “top off” a heap assembled from an even number of whole reams, e.g., two whole reams totaling 900 sheets plus four quires of 25 sheets each for an edition of a thousand copies. In a two-press shop such as Orwin's, concurrent printing on two presses increased both the complexity of the warehouse keeper's job and the possibility of the mixing of papers from different job-lots. In general, inferences about which method was used to build the heap are limited by the “tip of the iceberg” evidential situation noted earlier. Even a relatively large sample of an edition, e.g., twenty-one books, forty-five books, etc., may exhibit but a single example of an anomalous paper. However, the fact that the warehouse keeper built the heap from units of quires suggests that at least twenty-four other examples of that paper occurred in the whole edition. But that possibility is by no means a certainty. The warehouse keeper's duties included, in Moxon's terms, the process of “culling the chording quires” in search of “good” sheets. He held each sheet up to the light to determine whether it was totally useless or merely exhibited minor flaws which would not be noticeable in the interior gatherings of a book or after cropping once the book was bound. The usable sheets were then assembled into quires and added to the heap for the current job, or set aside until a complete quire was assembled. Since the chording quires contained a mixture of papers, it is possible that a single anomalous sheet is the only example of that paper in an edition.

Several examples will be helpful in illustrating the resulting phenomena. In the case of the Jonson Folio of 1616, James A. Riddell has demonstrated that the small papers used in some of the final quires were used in printing the preliminaries, thereby establishing that sequence. Similarly, his charting of the watermarks in a large number of copies shows that, contrary to the accepted view, the printing of each sheet did not begin with large paper and finish on small paper, but rather, the next day's printing began on the same size as that which concluded the previous day's printing.[14] The two job-lots of paper (i.e., small and large papers) were clearly ordered upon the basis of accurate calculations which permitted the warehouse keeper to build the heap so that the desired proportions of small and large copies would result. He probably built the heap sequentially from alternating numbers of reams of large and small paper. George Eld's printing of Jonson's Sejanus (STC14782, 1605) represents another instance of two separate job-lots of paper being used in a single edition. Although the watermarks in only thirteen copies have been examined, a consistent distribution pattern obtains in which each book

contains papers exclusively from one of the job-lots. One lot consists of inexpensive pott paper bearing the initials “BC” topped by a cluster of grapes. The other is a more expensive grade in two varieties of watermarks, the first consisting of the triple set of initials “IR AR| IP” topped by a crown-like design, the second a single “IAR” initial. Again, the warehouse keeper appears to have built the heap to specification.[15]

The processing by the warehouse keeper produced distributions which can be useful in demonstrating concurrent production on two presses. For example, in his seminal study “New Uses of Watermarks as Bibliographical Evidence,” Allan Stevenson examined the watermark evidence found in three plays printed by Thomas Cotes in 1639/40—The Night-Walker, The Opportunitie, and The Coronation—in demonstrating that the latter two had to have been printed concurrently from a single job-lot of paper. In this instance, the sequence of watermarks through the books is virtually identical (see discussion 155-165).[16]

Similarly, watermarks can demonstrate the sequential printing of two editions of one work or two separate works. In this case, the overlapping of papers provides the evidence, that is, varieties found in the final gatherings of the first edition appear in the early gatherings of the next edition. The watermarks in the first two editions of John Marston's The Malcontent Q1-2

(STC17479, 17480; 1604), for example, provide evidence both of sequence and concurrent shared printing in two shops. In both editions, primary printer Valentine Simmes printed F-H and A, while George Eld printed B-E. The sequence of papers in Q1 (a dozen extant copies) shows that both Eld's and Simmes's presses were fed from the same job-lot of paper on a daily basis. Furthermore, the sequence of editions exhibits an overlapping of papers. The watermark that first appears in the final sheet of Q1 printed by each printer also appears in their respective first sheets of Q2 (six extant copies).

Finally, watermark evidence can in some instances either confirm or disprove external dating evidence such as that found in Stationers' Register entries, imprints, and dated dedications. For example, all three editions of The Malcontent bear the imprint date of 1604. However, correspondences between papers in the third edition and those found in proximate books in early 1605 demonstrate that the edition was printed in March 1605, and not in the period framed by the 5 July 1604 entry into the Stationers' Register.[17] Similarly, both editions of Robert Dallington's A Survey (STC6200, 6201) bear a 1605 imprint date and specify George Eld as the printer. The supposed first edition (STC6200) contains papers used in another 1607 book printed by Eld (Richard Parkes, An Apology STC 19259), pointing to a 1607 printing with a false imprint date of 1605. In short, the watermark evidence reverses the sequence of editions, and places the actual second edition (STC 6200) two years later than the imprint date of 1605.[18]

These two cases are useful for demonstrating the difference between positive and negative inferences based upon paper evidence. In both cases, the inference can be stated negatively, e.g., The Malcontent Q3 was not printed in the 5 July 1604 period, A Survey Q1 was not printed in 1605. However, these negative inferences do not result from the paper evidence directly, e.g., since the papers from the two respective earlier dates do not appear, the books were not printed then. Rather, they are the opposite of the positive inferences which the paper evidence does prove, e.g. The Malcontent Q3 was printed in early 1605, A Survey Q1 was printed in 1607. In the absence of the papers which establish the later printing dates, no inference about dates could be drawn.

Before proceeding to an analysis of the watermark evidence, it should be noted that the books surveyed in this study are divided into two groups for the sake of expedience. External evidence (eleven Stationers' Register entries and two dated dedications) permits a tentative dating of thirteen of the twenty-nine books printed by Orwin in 1591. This includes Daphnaida with its dedicatory date of “this first of Januarie. 1591.” Although the inclusion of Daphnaida in this group may seem illogical because its date is exactly what is in question, the fact is that every external date associated with the books in

a paper survey must be considered tentative until the paper evidence demonstrates otherwise. As noted immediately above, the paper evidence can confirm the external dates in some cases, and disprove it in others. In many instances, the paper evidence does not contradict the external date, so we accept the date in good faith. In regard to the date of Complaints, the evidence is a bit more compelling than the “in good faith” situation. The “1590” imprint date in the title-page of Muiopotmos on T1 and the Stationers' Register entry date of 30 December 1590 can only mean the year 1590, regardless of Old Style or New Style dating. The “1591.” dates on the remaining three title-pages must, therefore, mean 1591 New Style—otherwise, a totally untenable proposition emerges, e.g., that the other three sections were printed a year later in January 1591 Old Style (1592 New Style) using exactly the same job-lot of papers.

To return to the main issue, a statistical summary of the survey will provide a context for assessing the paper evidence which is relevant to the dating of Daphnaida. Nine of the thirteen tentatively datable books were examined. Two of the remaining four of those thirteen were entered early in the year (STC20588: 11 January, 2 sheets; STC22685: 4 February, 21 sheets) and their unavailability has created an unfortunate gap in the sense that the opportunity to stumble serendipitously across overlapping papers was eliminated. Even so, the survey includes the papers in seven tentatively datable books printed through May 1591 (113 edition-sheets); a total of 201 edition-sheets were identified in the nine tentatively datable books; seventeen additional sheets in second copies of Ivychurch STC11340 and Emmanuel STC11338.5 along with 115 sheets in five Folger copies of Complaints raised the total number of actual sheets to 333 in datable books. In addition, the survey included eight undatable books totaling 143 edition-sheets. Overall, the watermarks of 344 edition-sheets (476 actual sheets) were catalogued for the seventeen books. The twelve unexamined books contribute another 84 edition-sheets to the known output from Orwin's two-press shop in 1591. The sample consisting of 344 catalogued edition-sheets thus represents approximately 80% of the total output of 428 edition-sheets.[19]

It can be noted that, in the context of the dating issue, the survey of books other than Complaints and Daphnaida seems, in a sense, gratuitous since it could never prove that these papers were used only in those two books. However, a contextual survey is desirable because a negative inference is never the objective of a paper survey. In practical terms, a survey often enough yields evidence leading to a positive inference about the temporal proximity of the printing of two or more books, especially in a situation like this.[20]

Since active publishers like Ponsonby often ordered enough paper for an edition directly from a paper merchant rather than relinquishing this aspect of quality-control to the printer, the three books entered by Ponsonby early in the year were of special interest: The Countesses of Pembroke's Emmanuel STC11338.5 and The Countesse of Pembroke's Ivychurch STC11340, both entered 9 February 1591; and Complaints STC23078, entered 29 December 1590, but bearing both 1590 and 1591 imprint dates.[21] Along with Daphnaida, these four books constitute the “Ponsonby's books” subset of all the books printed by Orwin in 1591. In some such instances, the papers exhibit correspondences which indicate, at the least, delivery from a single London paper merchant, perhaps as a single order assembled at the warehouse from a single continental shipment. The watermark evidence found in Complaints and Daphnaida constitutes such a case.

The accompanying chart shows the watermark distributions for three copies of Complaints and one copy of Daphnaida. Although Muiopotmos is the fourth section of Complaints, it appears as the first section to be printed, anticipating the typographical evidence discussed later. Two families of

watermarks (as explained in note 8) appear. The “cuffed hand [or glove] pointing at star” designs are assigned identification numbers, while the “initialed pot [or vase]” designs are identified by the inscribed initials, or in instances where the initials are undecipherable, simply by “P2” (i.e., “pot 2”) and “P3”. Although five Folger copies were examined, only copies 1 and 2 are shown: time restrictions prevented accurately identifying the “hand-star” papers found in all sheets of Folger copy 3 except for A (where the “IB” pot appears), and those appearing in the mixtures in Folger copies 4 and 5.

Watermark Distribution in Spenser's Complaints and Daphnaida

                                             

Muiopotmos
Sheet	T	V	X	Y	Z
Huntington	1	2	3	2	4
Folger copy 1	6	5	3	6	6
Folger copy 2	RB	7	RB	P3	GG
Complaints
Sheet	A	B	C	D
Huntington	IB	5	6	7
Folger copy 1	IB	6	6	5
Folger copy 2	IB	IB	IB	IB
Tears of the Muses
Sheet	E	F	G	H	I	K
Huntington	5	6	6	6	1	6
Folger copy 1	6	6	5	6	1	6
Folger copy 2	P2	5	RB	RB	10?	P3
Prosopopoia
Sheet	L	M	N	O	P	Q	R	S
Huntington	6	1	6	5	8	3	6	9
Folger copy 1	5	1	6	5	8	5	4	7
Folger copy 2	RB,P2	RB	IB	RB	RB	4,GG	RB	RB
Daphnaida
Sheet	A	B	C
Huntington	1	6	5

As can be noted in the chart, nine distinct watermark designs of the “hand pointing at star” family can be identified.[22] More may be present; at times the condition of a watermark prevents taking the precise measurements needed for its identification, and in other cases, portions of the design are distorted or obscured because of various factors. In general, the nine designs are very similar and hence difficult to distinguish in marks 1, 2, 3, 5, and 6; marks 4, 7, and 8 are characterized by a very deep hollow where the thumb joins the palm; and mark 9 is distinguished by widely spaced fingers. In regard to the second family of “initialed pot/vase” watermarks, the progress of this study provides an illuminating example of the principle noted earlier, namely, no final conclusion can ever be reached regarding the actual number of different papers used in a book unless every book in the edition could be

examined, an impossibility. Prior to examining the Folger copies, the following sentences appeared at this point in an earlier draft of this study:

As the chart shows, the “IB” “initialed pot/vase” design does appear in other copies, and furthermore, four other vase designs supplement the dominant “hand-star” watermarks. So, examining the remaining sixty-nine copies could produce yet more watermarks as well as modifying the apparent pattern of distribution. At this point, the evidence from just six books yields three copies (Huntington, Folger copies 1 and 3) which exclusively use the “hand-star” family of watermarks except in sheet A, two copies which exhibit a roughly even mixture of “hand-star” and vase papers (Folger copies 4 and 5), and one copy (Folger 2) which contains a large majority of vase papers with only three “hand-star” papers in the remaining sheets. In four of the copies, the “IB” vase can be identified in sheet A, and probably is the same seen in the other two exemplars of sheet A. In short, the six copies exhibit an unbroken pattern of the “IB” or other pot/vases in sheet A. The distribution patterns in the remaining sheets consist either of exclusive “hand-star” papers, or mixtures in varying proportions with vase papers. The sample is simply not large enough to draw any inferences about the composition of the papers in the job-lot other than to conclude that at least two families of papers were included, and in some copies, one family was used exclusively except in sheet A.

Three anomalies occur. In the Folger copy 3 (all “hand-star”), sheet A is composed of two half-sheets, with the “IB” vase appearing in A2/3. A totally unrelated paper appears in A1/4 bearing a largely undecipherable inscription (possibly “L[u,e?]nto Mor[o?]” which has a tantalizing ring to it) of the kind that is frequently found in larger sizes of paper. Next, the first gathering of section 3 (Prosopopoia, sheet L) in the Folger copy 2 is made up of one half-sheet exhibiting the “P2” watermark in L2/3. However, the base of the “RB” vase appears in L1, but the crest is not discernible in L4. Two factors deflect the suspicion that L1 and L4 are not conjugate: (1) although the chain lines do not match exactly in the gutter of L1/L4, the spacing of the chain lines as a whole is erratic, and the variations in spacings could be attributable to the length of the paper that is buried in the binding gutter; (2) in two other instances (O1/O4 and S2/S3), the crest and base respectively appear without the other half of the watermark, but chain line measurements coincide with conjugacy nonetheless. So, in the final analysis, L4 may be conjugate with L1. A third anomaly occurs in Folger copy 2. Sig. Q1 exhibits the fingers of the “hand-star” watermark #4, but is not conjugate with Q4; and Q2/3 are conjugate and exhibit the “GG” vase watermark. The rest of the sheets in Folger copy 2 are whole. As is usual, it is difficult to define the exact reason for these

anomalies: numerous possibilities have been inferred in other studies. Nonetheless, they do demonstrate that an effort was made to salvage the good portions of damaged sheets. Again, the sample does not allow inferring that this occurred only in the context of copies containing mixtures of the two families of watermarks.

To return to the central question, the evidence does demonstrate that the Huntington copy of Daphnaida was printed on three papers identical to those found in the complete Complaints (i.e., all four sections), or, in other words, on papers from the same job-lot. Given the dedication date three days after the entry of Complaints and the sharing of the papers, the conclusion that Daphnaida was printed in close temporal proximity with Complaints is inescapable.

Furthermore, the paper evidence found in the two other books printed by Orwin for Ponsonby in early 1591 indicates that they also were temporally proximate to Complaints and Daphnaida. Papers identical to those in Complaints and Daphnaida appear in Ponsonby's next two books, Ivychurch (copy 1 = Huntington shelfmark 59822; copy 2 = 59746) and Emmanuel (copy 1 = 59823; copy 2 = 59747) in the following sheets:

             

Ivychurch (copies 1 and 2)	Emmanuel (copies 1 and 2)
#1: A, C
#3: D, G	#3: B
#4: D, I, M	#4: C
#5: L
#6: A, K
#9: M	#9: M

Three additional “hand-star” designs appear in Ivychurch and Emmanuel, yielding a total of nine watermarks which appear in these two books, six of which also appear in Complaints and two in Daphnaida. Overall, thirteen different identifiable “hand-star” watermarks appear in the four books. The sharing of identical papers in the four books leaves no doubt whatever that a single job-lot of paper containing the “hand-star” papers was used in the printing of the four books.[23] The conclusiveness of the evidence cannot be overstated. Further, it should be noted that the job-lot included a supply of vase papers that may also have been used in Ivychurch and Emmanuel, although no appearances were found in the four copies that were examined.

It is reasonable to assume that, if a single job-lot of papers was purchased at one time for use in a sequence of four books, their printing had to be more or less temporally sequential—for instance, it would make no sense to claim that a publisher purchased a job-lot of papers that would be used in one book in January, two other books in February, then a fourth book in January of the next year. In this instance, the Stationers' Register entry dates of 30 December 1590 and February 1591 make it a virtual certainty that the four books were printed in close temporal proximity from one job-lot of paper. This inference can be confirmed by a reconstruction of the printing context of the proximate period.

Reconstructing the production schedule of a specific period is, at best, a tentative process unless either external or internal evidence suggests and/or confirms the probable sequence and locations of all books produced during that period. Even then, the key variable is unknown, namely, the edition-size (i.e., number of copies printed) of each book, and in turn, the absence of that crucial fact forces the bibliographer to resort to inexact mathematical techniques such as deriving averages for weekly production rates based upon the known annual output from a shop. However, the averaging technique has the advantage of leveling the short-term variations in production that could arise from any number of causes. For example, periods of very low production or total inactivity could result from the lack of sufficient work to keep Orwin's two presses in full operation. In addition, since the early hand-press was prone to various problems, work stoppages were a matter of course. In terms of edition-size, it was possible to print and perfect two edition-sheets of an edition of 500 copies in one work-day, given the usual daily production rate of 1250 perfected sheets at the press. Whether this rate could be achieved in small editions depended entirely upon the rate at which the compositor could set type—it was virtually impossible if the text consisted of packed prose, but perhaps within reach in verse texts set in larger type such as the 94mm roman in stanzaic form with a lot of “white space” on each page, the kind of setting situation seen in Complaints and Daphnaida. On the other hand, an edition-sheet in a large edition of 1250 or 1500 would have inevitably required at least one day for printing and perfecting. So, in absence of exhaustive records such as were available for McKenzie's study of the Cambridge University Press,[24] it is virtually impossible to arrive at anything better than average figures about production rates during a specific period. Nonetheless, such figures are useful in determining whether, given the average production rate, a particular sequence of books could have been printed during a period. In the present case, the question is: do the books that can be assigned to the January-February period represent a “normal” rate of production as suggested by the weekly average? If so, it is reasonable to infer that the sequence was printed during the period as suggested by the external and internal dating evidence.

Entry dates in the Stationers' Register locate six books in the period, and

the watermark evidence demonstrates that Daphnaida belongs there as well. As noted earlier, the edition-sheet is the unit of production in terms of the output from the press. Although slightly less than half of the books (14/29) produced by Orwin in 1591 are datable, these seven books containing 72 edition-sheets can be assigned to the January-February period, and represent 17% of Orwin's known output of 428 edition-sheets for the year:

Complaints STC23078, 30 December 1590, 23 sheets
Daphnaida STC23079, 1 January 1591, 3 sheets
A Consort of Creatures STC20588, 11 January 1591, 2 sheets [not checked]
The Harmony of the Church STC7199, 1 February 1591, 6 sheets
A Preparative to Marriage STC22685, 4 February 1591, 21 sheets [not checked]
Emmanuel STC11338.5, 9 February 1591, 5 sheets
Ivychurch STC11340, 9 February 1591, 12 sheets

Given the fact that Orwin owned two presses, the average production rates can be calculated on the basis of the known annual output and the six-day work-week totaling about 300 days per year (365 days less 52 Sundays and the two weeks of vacation during Christmas and Easter holidays). For 428 sheets and two presses, then, Orwin's production rate amounts to 8.4 sheets per week, and probably slightly more than half of that for a single press at full operation (two workers—pressman and beater).[25] If the 28 sheets of Complaints, Daphnaida, and Consort were printed during January, the production rate figures to roughly 7 sheets per week. Assuming that the edition sizes were such as required a full day at the press for each sheet, a single press could have done the job with several sheets spilling over into February, but the annual average production rate suggests that Orwin, on the whole, did not employ the single press production method. Thus, given the average weekly production rate for the year, it seems clear that Orwin's two presses were concurrently in operation during the period according to Orwin's probable production method.

As implied earlier, a reconstruction of the printing schedule requires terminal dates for the given period. In this case, the four books entered in early February provide the terminus for January, but no subsequent books were entered until 30 April. The printing of the 44 sheets in those four books obviously did not require three months. Similarly, the 51 sheets of the 30 April entry (STC11097) and the 13 sheets of the 12 May entry (STC21057) did not keep Orwin's two presses busy until the next entry on 3 August (STC5590). The 82 sheets of that book required about 10 weeks at the average rate, but the next entry (STC24913) occurs on 8 November, some thirteen weeks later. Overall, 224 sheets occur in the tentatively datable books, or 52% of the total output of 428 sheets. The remaining undatable 48% of the sheets (including one long book of 78 sheets, STC25626), then, would have been printed in the “gaps.” Most of the undatable books are short and of the

kind that could be “squeezed” into the schedule. Regardless, it seems clear that the dates of the January-February books correspond to what could be reasonably expected of concurrent production on two presses at the average weekly rate of 8.4 sheets.

In short, Ponsonby's four books were in all probability printed in the sequence indicated by the dates in a nearly continuous operation of the kind that characterized that method of production. Ponsonby apparently planned the four books as a single project and purchased the job-lot of paper in advance. The end result is the extensive sharing of papers among the four books.